Astronomy & Astrophysics manuscript no.

8322main February 2, 2008

(DOI: will be inserted by hand later)

Effects of photophoresis on the evolution

of transitional circumstellar disks
Fabian Herrmann1 and Alexander V. Krivov1
arXiv:0711.2595v1 [astro-ph] 16 Nov 2007

Astrophysikalisches Institut, Friedrich-Schiller-Universität Jena, Schillergäßchen 2–3, 07745 Jena, Germany

Received July 20, 2007, accepted 1 October 2007

Abstract. Although known for almost a century, the photophoretic force has only recently been considered in astrophysical
context for the first time. In our work, we have examined the effect of photophoresis, acting together with stellar gravity,
radiation pressure, and gas drag, on the evolution of solids in transitional circumstellar disks. We have applied our calculations to
four different systems: the disks of HR 4796A and HD 141569A, which are several Myr-old AB-type stars, and two hypothetical
systems that correspond to the solar nebula after disk dispersal has progressed sufficiently for the disk to become optically thin.
Our results suggest that solid objects migrate inward or outward, until they reach a certain size-dependent stability distance
from the star. The larger the bodies, the closer to the star they tend to accumulate. Photophoresis increases the stability radii,
moving objects to larger distances. What is more, photophoresis may cause formation of a belt of objects, but only in a certain
range of sizes and only around low-luminosity stars. The effects of photophoresis are noticeable in the size range from several
micrometers to several centimeters (for older transitional disks) or even several meters (for younger, more gaseous, ones). We
argue that due to gas damping, rotation does not substantially inhibit photophoresis.

Key words. planetary systems: formation – planetary systems: protoplanetary disks – circumstellar matter – celestial mechanics
– stars: individual: HR 4796A, HD 141569A.

1. Introduction
The standard (core accretion) scenario of planet formation
implies continuous growth of solids in a protoplanetary disk
around a young star (see, e.g. Safronov 1972; Wetherill
1980; Shu et al. 1987; Lissauer 1993; Strom & Edwards
1993; Weidenschilling & Cuzzi 1993; Papaloizou et al. 1999;
Blum & Wurm 2000; Wurm & Blum 2006; Henning et al.
2006; Meyer et al. 2007). Micron-sized dust grows step by
step to larger bodies, until asteroid-sized planetesimals are
reached. At sizes well below 1 km, gravitational interac-
tions between the objects play a minor role and the pro-
cess is largely determined by interactions of solids with the
ambient gas in the disk. Gas causes sedimentation, mix-
ing, radial drift (Weidenschilling 1977; Kley et al. 1993;
Takeuchi & Artymowicz 2001; Brauer et al. 2007) and other
effects that all dictate the spatial, size, and velocity distribu-
tions of grains and therefore the conditions for their growth.
As the disk evolves, both gas and dust are gradually removed
from the systems. In this process, the gas-to-dust ratio re-
duces from ∼ 100 in the initial protoplanetary disk stage, at
the ages on the order of 1 Myr, to vanishing values of ≪ 1
at the debris disk stage after ∼ 10 Myr (Lawson et al. 2004;
Haisch et al. 2005; Hollenbach et al. 2005; Takeuchi et al.
Send offprint requests to: Fabian Herrmann, e-mail:
fabian@astro.uni-jena.de
2005; Jayawardhana et al. 2006; Balog et al. 2007; Currie et al.
2007).
Transition from gas- and dust-rich, optically-thick pro-
toplanetary disks to nearly gas-free, optically-thin debris
disks is currently in the focus of interest of both observa-
tional and theoretical effort (Wieneke & Clayton 1983; Morfill
1983, 1988; Weidenschilling & Cuzzi 1993; Strom et al.
1993; Zuckerman & Becklin 1993; Simon & Prato 1995;
Meyer & Beckwith 2000; Ardila et al. 2005; Calvet et al. 2005;
Hueso & Guillot 2005; Augereau 2006; Bouwman et al. 2006;
Eisner et al. 2006). Examples of transitional objects are TW
Hya, HR 4796A and HD 141569A, while the well-known sys-
tem β Pic already enters into the realm of almost gas-free de-
bris disks (Thébault & Augereau 2005). Resolved transitional
disks exhibit radial structure in the form of rings and gaps, of-
ten alternating, and there is a lot of debate whether this struc-
ture is caused by gravity of hidden planets (Augereau et al.
1999b; Wyatt et al. 1999; Telesco et al. 2000) or by interac-
tion of solids with the ambient gas component (Klahr & Lin
2000; Takeuchi & Artymowicz 2001; Besla & Wu 2007). The
first possibility is supported by the fact that the amount of gas
remaining in these systems is probably no longer sufficient to
form Jupiter’s gas envelope (Chen & Kamp 2004) and so, the
planet formation process must already be finished. The sec-
ond hypothesis is substantiated by simulations based on the
observational estimates of the gas contents. These show that
gas drag, acting together with other forces – stellar gravity and
2 Herrmann & Krivov: Photophoresis in transitional disks
radiation pressure — could result in segregation of different-
sized solids and radial fractionation of dust.
In this paper, we study the motion of different-sized solids
under the combined action of stellar gravity, radiation pressure,
and gas drag, to which we add photophoresis, an additional
radial force that acts on particles in a gas disk which are ex-
posed to the radiation field of the central star. Photophoresis,
one of the lesser known forces of physics, has been first de-
scribed by Ehrenhaft (1917). The force is caused by the fol-
lowing process. If an object is embedded in a thin gas and
is exposed to an anisotropic radiation field (i.e. of a star), a
gas molecule that becomes accommodated by the object’s sur-
face and rejected again will depart from the illuminated – and
therefore warmer – side on average with a greater velocity
then from the dark, colder one. Net momentum is transferred
to the object, accelerating it away from the light source (note
that for very small particles, the force can also be attractive –
see Tehranian et al. 2001). Although known for almost a hun-
dred years and successfully used in technical applications, such
as the construction of optical traps (Steinbach et al. 2004), it
has only recently been analyzed in astrophysical context. The
first attempt to examine its influence on planetary formation
was made by Krauss & Wurm (2005). In their subsequent pa-
per (Wurm & Krauss 2006), they investigated how photophore-
sis might affect the formation of chondrules and the survival
of Calcium/Aluminum-rich inclusions. Finally, Krauss et al.
(2006) analyzed the effects imposed by photophoresis on the
inner rim of a dusty protoplanetary disk. They assumed the disk
to stay optically thick at all times, thus allowing only its in-
ner edge to be influenced by photophoresis. Using an exponen-
tial law to describe particle growth and the α 1+1D turbulent
model (Shakura & Sunyaev 1973) used in Alibert et al. (2005)
and Papaloizou & Terquem (1999) for the evolution of the gas
disk, Krauss and Wurm computed the motion of the inner edge.
This is identical to the motion of the smallest particles, since,
as will be shown subsequently, for solar luminosity systems,
the velocity of outward motion of an object increases with its
size in a certain size range (compare Fig. 3). In essence, they
showed that the dust disk’s edge first moves outward, pushed
by photophoresis. After a few million years of time, it shrinks
towards the star again, because gas pressure declines as disk
dispersal proceeds.
This paper focuses on transitional disks. In these systems,
the conditions are quite favorable for photophoresis to be effi-
cient. Transitional disks are already optically thin, so that the
particles are exposed to the strong radiation field. On the other
hand, they are still sufficiently gas-rich. We would like to check
whether, and to which extent, photophoresis may affect the ra-
dial migration of solids.
We use a single-body dynamics approach to investigate
the behavior of a two-dimensional swarm of particles, the
sizes of which are distributed in an interval ranging from
several micrometers to several meters. The size of an indi-
vidual particle, though, is being kept constant. We develop
a simple theory for the long-term behavior of a distribu-
tion of solids experiencing gravity, radiation pressure, gas
drag and photophoresis. Methodically, our study is similar to
Takeuchi & Artymowicz (2001), whose computations we gen-
eralize to the presence of photophoresis. To solve the equa-
tion of motion, we employ a modification of the semianalytic
scheme used by Weidenschilling (1977). The results are ap-
plied to circumstellar disks with different gas density around
stars of different luminosity.
In Sect. 2, we discuss the astrophysical status and parame-
ters of the systems chosen for this study, as well as the formu-
las used for gas density and temperature in the disks. Section 3
provides formulas for different perturbing forces, including the
photophoretic one. Section 4 deals with the equation of motion.
In Sect. 5, we analyze the properties of the particles’ radial mo-
tion and derive a formula for the radii of stable orbits. We also
estimate the size ranges in which photophoresis should be taken
into account. In Sect. 6, we examine the problem of particle ro-
tation in order to check whether it might nullify the effect of
photophoresis. In Sect. 7, the results are summarized and ideas
for future research are presented.
2. The systems
Following Takeuchi & Artymowicz (2001), we choose two
transitional disks of particular interest for our study: HR 4796A
and HD 141569A. We use these two systems in our work as
model environments for the forces we wish to explore, as their
properties are quite well known and reasonably accurate mod-
els exist for the physical conditions (gas density and tempera-
ture) in their disks.
The A0-type star HR 4796A, located at a distance
of 67.1+3.5
−3.4 pc from the sun (Hipparcos data), is a mem-
ber of the TW-Hydrae-Association (TWA in the follow-
ing, see Kastner et al. 2001). According to current knowl-
edge, this is the young star association closest to our sun
(Zuckerman & Song 2004). The age of TWA can be estimated
to be approximately 10 Myr. Since it has used up its molecular
cloud completely, the only gas left in in the association is bound
in the circumstellar disks. The stellar parameters of HR 4796A
are estimated to be M⋆ = 2.5 M⊙ , L⋆ = 21.0 L⊙ (Koerner et al.
1998; Jayawardhana et al. 1998; Telesco et al. 2000). IRAS
discovered that it emits 0.5 % of its entire radiative power in the
infrared part of the spectrum, which points to a high dust den-
sity in its neighborhood – in fact, HR 4796A is the dust-richest
star in the Bright Star Catalog. Stauffer et al. (1995) estimate
its age to be 8 ± 2 Myr, which makes it slightly younger than β
Pic (12 Myr). The structure of the dust disk around HR 4796A
is highly complicated. First resolved images in thermal infrared
were obtained by Koerner et al. (1998) and Jayawardhana et al.
(1998). Wahhaj et al. (2005) conducted elaborate studies of
the disk’s structure, using data from MIRLIN at the Keck II
telescope as well as 350 µ m measurements from the Caltech
Submillimeter Observatory and Hubble Space Telescope scat-
tered light images. Their studies point to a disk composed of an
inner, exozodiacal dust ring located at roughly r = 4 AU from
the star and a wide, two-component outer dust belt consisting
of a broad ring of ∼ 7 µ m grains stretching from 45 to 125
AU, and a narrower structure between 66 and 80 AU consist-
ing of ∼ 50 µ m grains. While the exozodiacal dust may be the
product of an asteroid-type belt, the outer belt can be explained
most naturally by the assumption that the grains are emitted by
 Herrmann & Krivov: Photophoresis in transitional disks 3

Table 1. Parameters of gas disk models.

System Example ρ0 [10−10 kg m−3 ] Density exponent q Luminosity L⋆ /L⊙ Mass M⋆ /M⊙
lGlL ‘older’ SN 3.12 -2.75 1.0 1.0
hGlL ‘younger’ SN 156 -2.75 1.0 1.0
lGhL HR 4796A∗ 3.12 -2.75 21.0 2.5
hGhL HD 141569A∗ 156 -2.25 22.4 2.3

∗ Sources: Takeuchi & Artymowicz (2001), their Table 1

an exo-Kuiper swarm of planetesimals. Since the smallest ones
are diffused rapidly by radiation pressure, they form a broad
belt, while the larger ones stay closer to their area of origin.
Also, asymmetric structures have been reported (Telesco et al.
2000). They may be caused by secular perturbations of one or
several planets or by the stellar companion HR 4796B. Most
probably, the disk is perturbed by planetary bodies as well as
by the B component. In order to define the mass and number of
planets, the orbit of HR 4796B would have to be known with
greater precision.
The second object, HD 141569A, has the Hipparcos
distance of 99 ± 10 pc. Its parameters are estimated
as M⋆ = 2.3 M⊙ , L⋆ = 22.4 L⊙ (Jura et al. 1993, 1998;
van den Ancker et al. 1998) It is underluminous for its spectral
type B9.5 Ve, which is a common occurrence for young A-type
stars, found also at HR 4796A, β Pic and 49 Ceti (Jura et al.
1998; Lowrance et al. 2000). Being, like HR 4796A, part of
a multiple system with an M2 and an M4 component, the
lower-mass companions which have not yet reached the main
sequence can be used to find an estimate for the system’s age of
5 ± 3 Myr (Weinberger et al. 2000). As in the HR 4796A case,
the disk itself has a complex morphology (Weinberger et al.
1999; Augereau et al. 1999a; Fisher et al. 2000; Mouillet et al.
2001), consisting of two dust belts with radii of 200 and 325
AU, the centers of which are shifted by 20 – 30 AU in the
direction of the system’s semiminor axis. Between them a
dust-free gap can be found at ∼ 250 AU. Also, inside of 150
AU the disk’s luminosity declines to the level of background
noise, which points to a strong depletion of dust in the system’s
inner region. The outer ring shows a tightly-wrapped spiral
structure, which, according to the numerical calculations
of Augereau & Papaloizou (2004), can be produced by the
gravitative perturbation of HD 141569B and C. The dust gap
at 250 AU may be caused by a planet of approximately Jovian
mass, but it is not yet clear whether gas giants can form within
a few million years at a distance of several hundred AU from
the star (Wyatt 2005).
Both disks – HR 4796A and HD 141569A – surround lumi-
nous, early-type stars. Since we wish to explore the dependence
of the particle dynamics on the stellar luminosity as well, we
add two more, hypothetical systems with gas densities equal
to those of HR 4796A and HD 141569A, but around a star of
solar luminosity. They can be viewed as representations of the
solar nebula at different stages of dissipation (Hollenbach et al.
1994; Hollenbach et al. 2000).
For gas density and temperature in all four model systems,
we use standard power-law approximations (Hayashi et al.
1985):
 1/4 
T L⋆ r −1/2
= 278 (1)
1K L⊙ 1 AU
and
ρg  r q
= ρ 0 , (2)
1 kg m−3 1 AU
where ρ0 and q are constants that vary from one system to an-
other. To keep the treatment simple, we assume the exponent q
to be constant throughout each disk and do not use a ‘break-off
function’ to describe the density drop beyond a certain radius
rout (Takeuchi & Artymowicz 2001). As will be shown subse-
quently, the outer part of the radial profile does not really mat-
ter, because photophoresis affects only the inner parts of the
disk.
Altogether, our model contains three basic parameters: the
gas density ρ0 at r = 1 AU, the gas density exponent q which
controls the size of the gas disk, and the stellar luminosity L⋆ .
Our four model systems essentially explore the photophoretic
effect in the ‘parameter rectangle’ luminosity – gas density:
– lLhG (low luminosity, high gas content, corresponding to
the solar nebula at an earlier stage),
– lLlG (low luminosity, low gas content, solar nebula at a later
stage),
– hLhG (high luminosity, high gas content, HD 141569A),
and
– hLlG (high luminosity, low gas content, HR 4796A).
The parameters of all four systems are listed in Table 1.
3. Forces
In circumstellar nebulae, solid bodies experience a number of
perturbing forces in addition to gravity, causing them to move
along non-Keplerian orbits. In a first approximation, we can as-
sume that they describe circular orbits, the radii of which shrink
or grow in the course of time, depending on the size of the body
and its distance from the star. We now introduce the different
forces, presenting an analytic expression for each.
3.1. Photophoresis
While Rohatschek (1996) found a semi-empirical formula for
the photophoretic force at all gas pressures, Beresnev et al.
(1993) derived an analytic expression for spherical, nonrotating
4 Herrmann & Krivov: Photophoresis in transitional disks

objects with a homogenous surface using an elaborate theoret- hLlG (HR 4796A)
ical approach starting from the molecular velocity distribution 1e+12
function (compare also Tehranian et al. 2001). The acceleration
1e+10
due to the photophoretic force they found is:
1e+08
π µ g mH
r
I J1
aphot =
4 s ρbulk 2kT 1e+06

Kn
αE Ψ1 10000 1 mm
× . (3)
αE + 15ΛKn (1 − αE)/4 + αE ΛΨ2
100 1m

Here, s is the particle radius, I the radiation intensity, J1 the 1

asymmetry parameter that describes the accommodation of
gas molecules to the particle’s surface and light absorption. 0.01
0.1 1 10 100
Assuming complete absorption and an accommodation prob- r / AU
ability of 100 %, we set J1 = 0.5. The mean molecular weight hLhG (HD 141569A)
of a gas of solar composition is denoted by µg = 2.34 and the 1e+08
mass of a hydrogen atom by mH . Further, k is Boltzmann’s con- 1e+07
stant, T the gas temperature and ρbulk the density of the solid 1e+06
material. For icy aggregates, ρbulk ≈ 1000 kgm−3 is a good ap- 100000
proximation. The energy accommodation coefficient αE is the 10000
fraction of molecules in contact with the surface that accommo- 1000

Kn
1 mm
date to the local temperature, which enables them to contribute 100
to photophoresis. In our work we assume complete accommo- 10 1m

dation, and thus set αE = 1. The heat exchange parameter Λ, 1

which describes the particle’s thermal relaxation properties, is 0.1

determined as 0.01
0.001
kth + 4εσ T 3 s 0.1 1 10 100
Λ= , (4) r / AU
kgas
Fig. 1. Knudsen numbers Kn = L/s for systems with low (top)
where kth is the material’s thermal conductivity, ε its emissiv- and high (bottom) gas content as a function of distance from
ity, σ the Stefan-Boltzmann constant and kgas the thermal con- the star. In each panel, four lines correspond to the particle radii
ductivity of the gas. Assuming black bodies, we set ε = 1. The of 1 mm, 1 cm, 10cm, and 1 m. Since L only depends on gas
quantities Ψ1 and Ψ2 are given by density, Kn does not vary with stellar luminosity.
!
Kn 2π 1/2 Kn
Ψ1 = 1+ ,
Kn + (5π /18) 5 Kn2 + π 1/2 Kn + π /4
! Here, p = ngas kT is the gas pressure (assuming ideal gas). In
1.21 π 1/2 Kn
 
1 15 transitional disks, we can usually assume to be in the high
Ψ2 = + Kn 1− , (5)
2 4 100 Kn2 + π /4 Knudsen number regime, except for large objects in the inner-
most parts of the disks (Fig. 1). The simplified expression (6)
and Kn ispthe Knudsen number. It is defined as Kn ≡ L/s, where will allow us to find useful approximate solutions for the parti-
L = 1/( (32) π ngas rg2 ) is the molecule’s mean free path, with cles’ stability radii (i.e. the radii of stable circular orbits). For
rg ≈ 10−10 m being the radius of gas molecules and ngas their large particles (s ≥ 10 cm) and small distances from the star
number density. (r ≤ 1 AU), the more general expression (3) will be used in nu-
Expression Eq. (3) for the photophoretic acceleration is merical calculations.
valid for all Knudsen numbers. In the case of high Knudsen The terms ϒhc , ϒrad and ϒgas are related to three different
numbers (mean free paths of molecules are large compared to processes which reduce the photophoretic effect. The first term,
particle sizes), Eq. (3) reduces to (see Krauss & Wurm (2005), ϒhc , corresponds to the transport of thermal energy through the
their Eq. 1, and Beresnev et al. (1993), their Eq. 26) object – i.e. heat conductivity, which reduces the temperature
gradient, thus lowering the efficiency of photophoresis. The
I pJ1 second one, ϒrad , describes thermal radiation from the parti-
aphot = , (6)
4ρbulk (ϒhc + ϒrad + ϒgas ) cles’ surface. Since it is proportional to T 4 , much more energy
where is radiated from the warm than from the dark side, which re-
sults in a substantial reduction of the temperature gradient, and,
ϒhc = kth T , (7) therefore, of photophoresis. Obviously, this process gains effi-
ϒrad = 4σ T 4 ε s , (8) ciency with increasing temperature. Finally, the third term ϒgas
q describes heat conduction away from the object’s surface into
ϒgas = p 2kT /π µg mH s . (9) the surrounding gas. Since it also shows a weak dependence
 Herrmann & Krivov: Photophoresis in transitional disks 5

on temperature (∝ T 1/2 ), it causes a reduction of temperature
gradient too.
Krauss & Wurm (2005) and Wurm & Krauss (2006) use
only the first term ϒhc in the denominator of Eq. (6), which is a
good approximation for particles with s ≪ 1 mm and only in the
case of low luminosity (i.e. solar-type) systems (see Sect. 5.2).
In order to analyze the relative importance of the three ‘coun-
tereffects’ ϒhc , ϒrad and ϒgas , we plot them for HD 141569A as
functions of particle size (Fig. 2) at three different distances.
1000
100
10
1
with Qpr being the radiation pressure efficiency. We will make
use of the constant B later. The effective ‘photogravitational’
acceleration is then
Meff
agrav, eff = −G r. (12)
r3
In our work, we simply set Qpr = 1, assuming the solids to
be black bodies. The reason is that generalization to non-black
surfaces requires a change not only in the emissivity, but also in
the asymmetry parameter J1 . The latter is not known for real-
istic materials, and its determination requires complicated nu-
merical procedures (see, e.g. Mackowski 1989). Analytical
solutions are possible, but only in the cases of high or low ab-
sorption coefficients (see Arnold & Lewittes 1982).

0.1 3.3. Gas drag force

[W m-1]

0.01 Unlike the other three forces (gravity, radiation pressure, pho-
0.001 tophoresis), gas drag is not a radial force. Since it results from
1e-04 the momentum transferred to the body by molecules impinging
1e-05 on it as it moves through the gas, the force vector is antiparallel
Υhc to the relative velocity ∆v of the dust grain with respect to the
1e-06 Υrad
Υgas gas.
1e-07
100 1000 10000 100000 1e+06 To calculate gas drag, we first need a model for the motion
s / µm of the gas component of the circumstellar nebula. Since the
pressure gradient supports gas against stellar gravity, it travels
Fig. 2. Quantitative comparison of the three reduction pro-
on circular orbits with a sub-Keplerian speed
cesses for hLhG (HD 141569A) at r = 5, 10, 20 AU (from
vg = vK 1 − η ,
p
above for each set of lines). For the other systems the result (13)
looks qualitatively similar, with ϒgas being in the case of the
gas-poorer systems completely negligible for all sizes s. and angular velocity
Ωg = ΩK 1 − η ,
p
(14)
Figure 2 shows that for fairly small particles (s ≤ 1 mm),
where vKp= GM⋆ /r is the Keplerian circular velocity, ΩK =
p
heat conduction (ϒhc ) is the main reduction process. At larger
sizes (s ≥ 10 cm), radiation (ϒrad ) takes over. In the intermedi- vK /r = GM⋆ /r3 the corresponding angular velocity, and η
ate interval, both processes contribute to the reduction of pho- is the ratio of pressure gradient force to gravity:
tophoresis. At yet larger sizes (s ≥ 1 m), heat conduction from 1 dp
the surface into the surrounding gas (ϒgas ) comes into play, ap- η =− . (15)
rΩ2K ρg dr
proaching values similar to that of ϒhc . As lower gas pressure
dramatically reduces the importance of ϒgas , it is completely The η ratio can be rewritten as:
negligible for all object radii in the case of the gas-poor sys-
µg −1 L⋆ 1/4
   
1
tems hLlG (HR 4796A) and lLlG. η = 1.1 × 10−3 −q
2 2.34 L⊙
 −1 
M⋆ r  1/2  r 1/2
3.2. Radiation pressure × ≡E . (16)
M⊙ 1 AU 1 AU
Since radiation pressure is proportional to r−2 ,
it can be
If the motion of the particles is subsonic (∆v ≪ vT ≈ cS
taken into account by introducing an ‘effective stellar mass’
where cS is the speed of sound – this is always the case in the
(Burns et al. 1979):
transitional disks), the gas drag acceleration is given by (see
Meff = M⋆ (1 − β ) , (10) Takeuchi & Artymowicz 2001):
3ρg
where β stands for the ratio of accelerations due to radiation aD = − vT ∆v , (17)
4ρbulks
pressure and gravity:
where
arad 0.5738 Qpr L⋆ /L⊙
β= = 4

8kT
1
2 4
agrav ρbulk[g cm−3 ] s[µ m] M⋆ /M⊙ vT = = × hvtherm i (18)
 −1 3 π µ g mH 3
s
≡B , (11)
1 µm is 4/3 times the mean thermal velocity.
6 Herrmann & Krivov: Photophoresis in transitional disks

The reaction of particles to the gas drag force critically de- In the first case, the radial drift velocity vr ≡ dr/dt computes to
pends on their size. Small objects adjust their velocity instan- (see Weidenschilling 1977, sect. 4.1, 4.2):
taneously, they are swept along with the gas component. Large
vr, small = ts ∆g = Ts (β + χ − η ) rΩK . (25)
ones react only sluggishly to gas drag, taking a longer time to
change their velocity substantially. In order to create a quan- In the second case, the orbit decays at a rate
titative measure for the tendency of solids to be influenced by r 2 r ∆g (β + χ − η ) rΩK
gas drag, the stopping time is introduced. We denote the time vr, large = − ∆v = = . (26)
ts vK ts agrav Ts
needed for a particle injected into the gas to be slowed down to
e−1 times its initial velocity by ts = ∆v / aD . The dimensionless These two formulas can be combined into (see
stopping parameter Ts is defined by: Takeuchi & Artymowicz 2001, Sect. 3.3):
4ρbulksvK β +χ −η
Ts ≡ ts ΩK ≈ . (19) vr = vK . (27)
3ρg rvT Ts + Ts−1
The right-hand side approximation holds in the case of sub- In Fig. 3, we plot the absolute value of vr over s for r =
sonic motion. It renders the stopping time independent of the 10 AU, both with and without photophoresis. The points where
particle’s momentary velocity. vr = 0 are shifted to larger sizes by photophoresis. Note that for
larger sizes, radial velocities depend almost exclusively on gas
density since for them gas drag (η in Eq. 27) becomes the main
4. Equation of motion driving force of radial migration.
The equation of motion of a dust particle is:
d2 r No Photophoresis
= agrav, eff + aphot + aD . (20) 1000
dt 2 hLlG (HR 4796A)
hLhG (HD 141569A)
100
Being interested in the radial motion of the bodies, we now lLlG (later SN)
lLhG (earlier SN)
consider the radial component of the equation of motion (20). 10
In the reference frame corotating with the gas, it takes the form 1
|vr| / m s-1

d2 r 0.1
− Ω2g r = −agrav, eff + aphot . (21)
dt 2 0.01

The acceleration ≡ ∆g of the particle in the corotating

d2 r/dt 2 0.001
frame can be interpreted as the ‘residual gravity’. It computes
1e-04
to
1e-05
∆g = −agrav, eff + aphot + Ω2g r = (β + χ − η ) Ω2K r , (22) 100 1000 10000 100000 1e+06 1e+07 1e+08
s / µm
where χ ≡ aphot /agrav is the photophoresis-to-gravity ratio. Photophoresis
Particles experience the inward-directed residual acceleration 1000
hLlG (HR 4796A)
η Ω2K r and the outward-directed accelerations due to pho- 100
hLhG (HD 141569A)
lLlG (later SN)
tophoresis (χ Ω2K r) and radiation pressure (β Ω2K r). lLhG (earlier SN)
10
Assuming now that the solid particle moves at a Keplerian
circular speed, we can write 1
|vr| / m s-1

0.1
v2g v2
= K + ∆g . (23) 0.01
r r
The relative velocity is then approximately 0.001

∆g 1e-04
 
∆v = vK − vg ≈ − vK . (24) 1e-05
2 agrav 100 1000 10000 100000 1e+06 1e+07 1e+08
s / µm
We now use ∆g to derive an approximation for the ra-
dial drift velocity of the particles. We can discern two limiting Fig. 3. The absolute values of the radial velocities at r = 10 AU.
cases: The minima correspond to the sizes for which vr = 0.
– Small particles are swept along with the gas. They stay on
circular orbits, while moving with sub-Keplerian velocity, In what follows, differential equation (27) for r(t) will be
which causes them to experience the residual gravity pull studied analytically and solved numerically. In the realm of
∆g. large radial velocities (or equivalently, small particles), r(t)
– Large particles are decoupled from gas. They move on should be computed from the ‘exact’ equation of motion (20).
Keplerian orbits, experiencing a head wind that gradually However, our numerical tests have shown Eq. (27) to be accu-
reduces their angular momentum. rate enough for all particle sizes larger than ∼ 100 µ m.
 Herrmann & Krivov: Photophoresis in transitional disks 7

5. Results 5.2. Equilibrium distance

We now calculate the equilibrium distance which, according to
5.1. Radial motion Eq. (27), must satisfy
β + χ − η = 0. (28)
We used Eq. (27) to calculate the radial drift of different-sized
particles. What happens, in short, is that the particles migrate From this, an implicit expression for the radius rstab (s) of the
inward or outward, until they reach stable circular orbits on stable orbit can be derived:
which gravity, centripetal force (in the particle’s own inertial
v2K (rstab ) (1 − η (rstab ))
system), photophoresis and radiation pressure balance each rstab (s) = . (29)
other, thus permitting circular motion (vr = 0), while at the agrav, eff (rstab , s) − aphot(rstab , s)
same time ∆v = 0, i.e. the particle travels at the same speed
as the gas.
High luminosity

Phot. (low gas content)

No Phot. (low gas)
Phot. (high gas)
No Phot. (high gas)
Photophoresis 10
No Photophoresis
s = 500 µm

rstab / AU
10

1
r / AU

s = 1 cm s = 3 cm

1 s = 3 mm

100 1000 10000 100000 1e+06 1e+07

s / µm
Low luminosity

0 2000 4000 6000 8000 10000 12000 Phot. (low gas)

t / yr Phot. (high gas)
10 No Phot. (high + low gas)

Fig. 4. Distance as a function of time for particles of four dif-

ferent radii in the hLlG system (HR 4796A). Dashed lines:
rstab / AU

without photophoresis, solid: with photophoresis. The lower

1
border of the plot corresponds to the evaporation limit revap =
0.16 AU (defined by black-body equilibrium temperature T =
1500 K).
0.1

10 100 1000 10000 100000 1e+06 1e+07

As an example, Fig. 4 shows the r(t)-curves for particles
of four different sizes with and without photophoresis in the
system hLlG (HR 4796A). The particles start on circular or-
bits with r0 = 5 AU. In fact, initial circularity is not important,
because elliptic orbits are quickly circularized by the gas drag
force. The initial value r0 does not matter either, because the
particles migrate towards their equilibrium orbits within several
thousand years. While small, outward migrating bodies travel
slowly towards their equilibrium distance rstab (s), approach-
ing it asymptotically, the larger, inward migrating ones almost
‘drop’ onto their stability orbits, being stopped almost instanta-
neously. Note, however, that ‘instantaneous’ here refers to a de-
celeration process taking several centuries. Therefore, the large
particles travel essentially on Keplerian orbits, corresponding
to the second case described in sect. 4. With increasing particle
size, though, this process takes longer, as larger objects experi-
ence weaker drag acceleration than smaller ones. Another con-
clusion from Fig. 4 is that the stability radii are pushed outward
by photophoresis.
s / µm
Fig. 5. Stability radii as functions of particle radius s for two
high-luminosity systems (top) and two low-luminosity ones
(bottom). In the bottom panel, the ‘No Photophoresis’ curve
is identical for both systems, as they differ only in the value
of gas density, which does not affect the stability condition if
photophoresis is not taken into account (η only depends on the
density exponent q, not its absolute value – see Eq. 16). Again,
the lower borders of the plots correspond to the evaporation
limits.
The equilibrium distance rstab (s) depends on the system’s
parameters as well as on the particle size. Fig. 5 shows solu-
tions of Eq. (29), computed with and without photophoresis. It
can be clearly seen that photophoresis significantly increases
the radii of stable orbits – as expected: photophoresis pushes
particles away from the star. The curves exhibit a character-
istic shape: after branching off from the solutions computed
8 Herrmann & Krivov: Photophoresis in transitional disks
without photophoresis, they flatten. While for hLhG there is
only a small region in which the curve is flatter, the two low-
luminosity systems have well-developed plateaus, the length
and distance from the star of which increase with gas den-
sity. The reason for this is explained below. Note also, that
the plateaus in the rstab (s)-curve correspond to areas, where
solids have a higher surface density. At a certain particle size,
the curves drop to zero suddenly. This is a consequence of
mean free path becoming smaller than particle sizes – in this
area, Eq. (6) is not a good approximation anymore, and Eq. (3)
should be used instead to calculate the photophoretic accelera-
tion. Because in the low Knudsen number regime photophore-
sis decreases in strength as body size increases, Eq. (28) and
(29) do not have solutions above a certain body radius, which
causes rstab (s) to drop to zero, so that there are no stable orbits
for solids above this critical size. These considerations are visu-
alized in Fig. 6, where η and χ are plotted as functions of r for
different radii s. The third force parameter, β , can be neglected
for large particles (s ≫ 100 µ m), and therefore is not shown.
We can identify two cases. For small bodies, the χ (r)-curve
has two intersections with η (r). This means that in a certain
size range, (28)–(29) have got two solutions. In our work we
consider only the one farther out from the star, since in transi-
tional disks we expect most particles to drift inward from larger
radii, where they are produced by collisions between left-over
planetesimals in exo-Kuiper belts. Besides, for three out of four
systems (hGlL is the exception), the inner solution lies in the
sublimation zone. As s increases further, the χ (r)-curve moves
downward until it no longer crosses the η -curve, and thus no
solution of (28)–(29) remains. This means that above a certain
size no stable orbits exist.
1 tion given in Eq. (6). We need to calculate the following ratios:
η
χ (s = 1 mm)
χ (s = 3 cm)
0.1 χ (s = 30 cm)
evaporation limit
0.01
0.001
1e-04
1e-05
1e-06
0.01 0.1 1 10
r / AU
Fig. 6. The force ratios η and χ as functions of r for different
particle sizes, in the hLlG system (HR 4796A). The number of
intersections between the curves corresponds to the number of
solutions of (28)–(29).
Next, we determine how the exact shape of rstab (s) is influ-
enced by the choice of system parameters. Therefore, in Fig. 7,
we plot the stability radii for different choices of L⋆ , ρ0 and q.
Note that, when varying L⋆ in the bottom panel of Fig. 7, we
change M⋆ accordingly. Analyzing the plots in Fig. 7, we can
identify the following dependencies:
– Higher gas density ρ0 moves the curves’ middle parts to
larger r, which is, of course, a consequence of photophore-
sis being directly dependent on gas pressure. Also, the lim-
iting particle size beyond which no stable orbits exist de-
pends almost exclusively on ρ0 , while the influence of the
other parameters is rather weak.
– The density exponent q, which determines the slope of ρ (r),
also controls the slope of rstab (s): a lower absolute value
of q produces a steeper curve, as the gas disk spreads to a
greater distance from the star.
– Stellar luminosity L⋆ determines the characteristic shape
of the rstab (s) curve. While large luminosities produce a
monotonically decreasing curve, lower (i.e. solar) ones gen-
erate a plateau at a certain distance from the star, which
gives rise to a concentration of particles of different sizes
in that region.
The mechanism behind the third effect is to be found in the
denominator of Eq. (6). As we have seen in Fig. 2, in different
size ranges three different processes reduce the photophoretic
effect – heat conduction in small particles, radiation for larger
ones and heat conduction into the surrounding gas for very
large objects. If the stellar luminosity is low, radiation becomes
important only for relatively large sizes, while for smaller ob-
jects only ϒhc is relevant in Eq. (6). As ϒhc is independent of
size, the resulting curve tends to run parallel to the s-axis, cre-
ating the plateau.
In the following, we formulate these considerations quan-
titatively. We are going to compute solutions of Eq. (29)
for very large and very small particles, using the simplified
large-Knudsen approximation for the photophoretic accelera-
aphot, ϒhc
χ (ϒhc ) = (30)
agrav
and
aphot, ϒrad
χ (ϒrad ) = , (31)
agrav
where aphot, ϒhc and aphot, ϒrad denote the accelerations due to
(large Knudsen regime) photophoretic force using exclusively
ϒhc and ϒrad in the denominator, respectively. The first case is
a good approximation for small particle radii (s ≤ 1 mm, com-
pare Fig. 2), the second one for larger, but sub-meter objects
(1 mm ≤ s ≤ 1 m).
The quantities χ (ϒhc ) and χ (ϒrad ) can be computed from
the following expressions:

M⋆ −1 L⋆  µg −1
  
χ (ϒhc ) = 1.0 × 10 8
M⊙ L⊙ 2.34
−1 
ρd

r q
× −3
ρ0
1000 kgm 1 AU
 r q
≡ Z(ϒhc ) (32)
1 AU
and
 Herrmann & Krivov: Photophoresis in transitional disks 9

ρ0 = 3.12 ... 156 * 10-10 kg m-3, q = -2.25, L = 22.4 Lsun, M = 2.3 Msun It is easy to show that in the first (small-particle) case, Eq.
rstab
(29) can be solved for s:

high ρ0
sstab, small (r) B
= . (34)
1 µm 1/2
E (r [AU]) − Z(ϒhc ) (r [AU])q
10 low ρ0
For definition of B and E, see Eqs. (11) and (16). This allows
rstab / AU

us to calculate the size of particles on stable orbits at a certain

stellar distance r. Since q is negative, sstab, small tends to infinity
as r approaches zero. This produces the curve’s plateau, i.e. the
1 concentration belt. From Eq. (34), the belt’s radius computes
to:
 1/(q−1/2)
100 1000 10000 100000 1e+06
E
rbelt = (35)
s / µm Z(ϒhc )
ρ0 = 156 * 10-10 kg m-3, q = -1.5 ... -3.0, L = 22.4 Lsun, M = 2.3 Msun
For the four systems considered, the resulting distances are
rstab listed in Table 2.
small |q|

10
Table 2. Existence and parameters of the belt.
large |q|
rstab / AU

System rbelt /AU C

lLlG (‘older’ SN) 2.0 0.8
lLhG (‘younger’ SN) 6.5 2.6
hLlG (HR 4696A) 4.0 0.1
1
hLhG (HD 141569A) 22.8 0.07

100 1000 10000 100000 1e+06 In the second (large-particle) case, the corresponding ex-
s / µm pression
ρ0 = 156 * 10-10 kg m-3, q = -2.25, L = 1 ... 21 Lsun, M = 1.0 ... 2.3 Msun
sstab, large (r) B (r [AU])−1/2 + Z(ϒrad ) (r [AU])q+1
rstab = (36)
1 µm E
can be derived.
high L The condition for the formation of the belt is that at its
rstab / AU

position, the value of sstab, large has to be sufficiently larger

10
than sstab, small . If that is the case, the asymptotic behavior of
sstab, small is seen in the ‘exact’ curve too, otherwise it is over-
low L
ridden by sstab, large (see Fig. 8). To evaluate this condition nu-
merically, we set
′
rbelt = rbelt + ∆r (37)
10 100 1000 10000 100000
s / µm where for ∆r a sufficiently small value has to be chosen. We
use ∆r = 0.1 AU. This is necessary because sstab, small cannot be
Fig. 7. Dependence of the equilibrium distance rstab (s) on the
computed at rbelt , as the denominator of Eq. (34) becomes zero
gas density ρ0 , the density exponent q, and stellar luminosity
at this point. Then, the ratio of the two approximating functions
L⋆ . Top: ρ0 running, q and L⋆ fixed. Middle: q running, ρ0 and ′
at rbelt can be used as a measure of how pronounced the belt is:
L⋆ fixed. Bottom: L⋆ running, q and ρ0 fixed. The parameter
values are indicated on top of each panel. ′ ′
C ≡ sstab, large (rbelt ) / sstab, small (rbelt ). (38)
The values of C are listed in Table 2. Comparing with Fig. 5, we
find that for C ≪ 1, the plateau does not appear at all, or is only
marginal. For C ≈ 1, the belt is well developed. The particle
M⋆ −1 L⋆ 1/4  µg −1
  

size range it encompasses and its degree of concentration (i.e.
χ (ϒrad ) = 2.1 × 1010 the ’flatness’ of the curve in that area) increase with C, which
M⊙ L⊙ 2.34
−1  −1 makes it a direct measure for the system’s tendency to produce
ρd
 
r q+3/2 s
× ρ0 a belt. In fact, other choices for ∆r are possible – for them, the
1000 kgm−3 1 AU 1 µm values with which C has to be compared, have to be changed.
 r q+3/2  s −1 For ∆r = 0.1 AU, the critical value above which well-formed
≡ Z(ϒrad ) . (33)
1 AU 1 µm belts appear is C = 1.
10 Herrmann & Krivov: Photophoresis in transitional disks

100
hLlG (HR 4796A)
hLhG (HD 141569A)
lLlG (older SN)
lLhG (younger SN)
10
10
rstab / AU

rstab(s) (lLhG)
rstab small(s) (lLhG) 0.1
1 rstab large(s) (lLhG)
rstab(s) (hLhG)
rstab small(s) (hLhG)
rstab large(s) (hLhG)
0.01
10 100 1000 10000 100000 1e+06 1 10 100 1000 10000 100000 1e+06
s / µm s / µm

Fig. 8. ‘Exact’ function rstab (solid lines) and its two approx-
imations (dashed lines: Eq. 34; dotted: Eq. 36), for the lLhG
(thick lines) and the hLhG (thin) systems. Note how the be-
havior of rstab, small produces the curve’s plateau – in the higher
luminosity case, rstab, large ‘takes over’ at smaller particle sizes,
thus keeping rstab, small from producing the effect.
We see that low luminosities and high gas densities are cru-
cial for belt formation. Since ϒrad ∝ T 4 ∝ L⋆ , with rising lumi-
nosity ϒrad becomes important at smaller particle sizes, super-
seding ϒhc before the plateau of sstab, small is reached. Because
they increase the overall strength of photophoresis, higher gas
densities push the plateau outward, reducing its slope – and
thus increasing the degree of particle concentration in the belt.
Fig. 8 demonstrates how the exact solution of Eq. (29) is
approximated by Eq. (34) and Eq. (36) in different particle size
ranges for the lLhG and hLhG models.
5.3. Particle size range
As we have seen, particles of different sizes are influenced
by photophoresis to a different extent. Its relative importance
depends on their particle radius, declining for very small
(micrometer-range) and very large (above meter-range) ob-
jects. In between, photophoresis plays an important role. In
order to analyze the critical size range, we plot the ratio x ≡
|vr, phot /vr, no phot | as a function of s in Fig. 9, where vr, phot and
vr, no phot are the radial velocity with and without photophoresis
respectively. The distance is set to r = 10 AU. Note that the in-
terval between the two peaks corresponds to the size range in
which the two vr have got different signs, because the stability
radius with photophoresis is greater than 10 AU, while the one
calculated without photophoresis is smaller.
If we use the criterion that |x − 1| ≥ 0.001, photophore-
sis has to be taken into account in the size ranges listed in
Table 3. Note, however, that for s ≤ 10 µ m, the above evalu-
ation is no longer necessarily valid, as for small particles, the
photophoretic force can reverse its direction (Tehranian et al.
2001). Also, for hLhG (HD 141569A), β (smin ) is greater than
0.5, thus these particles are β -meteoroids which escape from
the system.
Fig. 9. Ratio of radial velocities, computed with and without
photophoresis. The stronger the deviation from unity, the more
important is photophoresis. The stellar distance is r = 10 AU
Table 3. Size range of particles affected by photophoresis.
System smin smax
lLlG (‘older’ SN) 9 µm 3 cm
lLhG (‘younger’ SN) 1 µm 1.5 m
hLlG (HR 4696A) 10 µ m 3.2 cm
hLhG (HD 141569A) 1 µm 6m
6. Rotation of particles
Until now, we considered only nonrotating particles. We have
to check whether this assumption is realistic, since the rotation
of bodies can transport thermal energy from the dark to the lit
side, thus banishing the photophoretic effect. There is a number
of different mechanisms that can induce particle rotation:
– Collisionally induced rotation. During the formation of
bodies as well as during their life, they collide with other
particles, which can change their orbits as well as transfer
angular momentum to them, spinning them up.
– Rotation induced by gas drag or radiation forces. If par-
ticles are not exactly spherical, any force acting on their
surface will change their angular momentum. For instance,
radiation pressure may spin-up the particles (the so-called
windmill effect, Paddack & Rhee 1975), but may also sta-
bilize/align then (Draine & Weingartner 1996). Such ef-
fects are beyond the scope of this paper.
Whether rotation is able to subdue photophoresis depends on
four different timescales: tcoll , the typical time between two col-
lisions; ts , the gas coupling time which also determines the time
needed to slow down rotation; theat , the thermal relaxation time
needed to establish a stable heat gradient within the particle;
and trot , the typical rotation period.
The collisional time tcoll can be estimated in a standard way:
1
tcoll = , (39)
np vσcoll
 Herrmann & Krivov: Photophoresis in transitional disks 11

where np is the particle number density, v the collisional veloc- 1e+06

High luminosity
ity and σcoll = π (s1 + s2 )2 is the collisional cross section for Low luminosity
spherical particles of radii s1 and s2 . With
ρd
100000
np = , (40)
ρbulk 43 π s3
10000
and the standard assumption ρd /ρg = 10−2, we get for particles
of equal size (cf. Krauss et al. 2006, their Eq. 15):
ρbulks 1000
tcoll = 33 . (41)
ρg v
The stopping time ts was computed in section 3.3, Eq. (19). 100
0.1 1 10 100
Finally, the thermal relaxation time theat is given by (see
v / m s-1
Krauss & Wurm 2005, Eq. 6):
1e-04
hLlG (HR 4796A)
ρbulkcd s2 hLhG (HD 141569A)
lLlG (later SN)
theat = , (42) lLhG (earlier SN)
kth 1e-05

where cd ≈ 1000 Wm−1K−1 is the particle’s heat capacity.

1e-06
We can use two different approaches to deal with rota-
tion. One is to compare collision time to stopping time: if
tcoll /ts ≫ 1, damping of rotation occurs faster than excitation by 1e-07
impacts, and therefore photophoresis is not significantly weak-
ened. Using Eqs. (41) and (19), we get 1e-08

tcoll vth
= 25 . (43)
ts v 1e-09
Another option is to check whether rotation, once it has been
excited by collisions, is sufficiently slow to allow a stable heat
gradient to be established within the particle: this is the case
if trot /theat ≫ 1. In order to find a rough estimate for trot , we
assume that the particle transforms the entire kinetic energy of
its radial motion vr into rotational energy:
1 2π 2
 
1
Jsphere = m v2r , (44)
2 trot 2 particle interactions as well as particle growth were not in-
with the solid sphere’s moment of inertia Jsphere = (2/5)ms2 ,
we get
8 πs
r
trot = . (45)
5 vr
Fig. 10 shows tcoll /ts as a function of v and trot /theat as
a function of s. The distance is chosen to be r = 10 AU.
We see that collisional timescales are longer than stopping
timescales by many decades, thus allowing rotation to decline
between collisions. Nonetheless, rotation frequencies can be
high enough to cancel out stable temperature gradients. We
conclude that photophoresis can be suppressed directly after
collisions through rapid rotation, but remains effective for most
of the time because gas drag damps spin faster than collisions
can excite it.
7. Conclusions and discussion
In this paper, we have analyzed the effect of photophoresis on
the dynamics of solid particles in the optically-thin, yet suffi-
ciently gas-rich, transitional disks around young stars. To this
100 1000 10000 100000 1e+06 1e+07 1e+08
s / µm
Fig. 10. Ratios of typical timescales. Top: tcoll /ts as a function
of v. Bottom: trot /theat as a function of s.
end, we employed a single-body approach and added the pho-
tophoretic force to the standard array of perturbing forces (stel-
lar gravity, direct radiation pressure, and gas drag). Particle-
cluded. We find that photophoresis may lead to noticeable cor-
rections to the results obtained with models that do not take it
into account.
Our main results can be summarized as follows:
1. Both with and without photophoresis, solid objects migrate
inward or outward, until they reach the stability distance
rstab (s), where s is the particle radius. At that distance, ra-
dial forces cancel each other in the particle’s own inertial
system, and the orbital velocity is equal to that of the gas.
The stability distance is a decreasing function of s, there-
fore particles are sorted according to size, with larger bod-
ies accumulating closer to the star. These results fully agree
with those by Takeuchi & Artymowicz (2001).
2. Photophoresis increases the stability radii, moving objects
to larger radial distances. The effect is noticeable in the
size range from several micrometers to several centime-
ters (for older transitional disks) or even several meters (for
younger, more gaseous, ones).
3. The steady-state distribution of solids is completely charac-
terized by the function rstab (s), the shape of which depends
on the system’s parameters:
12 Herrmann & Krivov: Photophoresis in transitional disks
– Higher gas densities move the curve to larger distances
without changing its overall shape. Gas density also
controls the maximum particle size up to which stable
orbits can exist.
– The steepness of the gas density radial profile deter-
mines the slope of rstab (s), with flatter profiles gener-
ating steeper curves.
– Stellar luminosity determines the curve’s shape. While
high luminosities (∼ 20 L⊙ ) produce simple decreasing
curves, low (solar) luminosities generate a plateau at a
distance from the star that can be computed with the aid
of Eq. (35). In this area, objects in a certain size range
may accumulate.
4. Particle rotation tends to reduce the photophoretic effect.
Our estimates suggest, however, that it is damped by gas
drag quickly enough to keep photophoresis at work.
Our work predicts the formation of a particle concentration
belt at a certain distance from the star. For the high luminos-
ity systems, it is not very pronounced (hLhG) or does not ap-
pear at all (hLlG). Furthermore, the radii of the observed rings
around HD 141569A are an order of magnitude larger than that
of the slight concentration belt predicted for the hLhG system.
It is not likely therefore that the observed structures around
HR 4796A and HD 141569A are caused by photophoresis.
While photophoresis is probably active in transitional disks,
the circumstellar rings of HR 4796A and HD 141569 must
be shaped by other forces and effects, such as gravitational
sculpting by planets or interactions with stellar companions.
Alternatively, a rapid decline of gas density at the disks’ edges
(Takeuchi & Artymowicz 2001) or a recently proposed dust-
gas instability (Besla & Wu 2007) may cause particles to accu-
mulate there.
The model presented here is rather exploratory and rests on
a number of simplifying assumptions. In the future, we plan to
investigate the problem a more realistic way, lifting some of
the assumptions we made to keep the problem tractable. First,
we plan to deal with particle-particle interactions, taking col-
lisions and growth into account. This can be done in the style
of Krauss et al. (2006), increasing the radius through an expo-
nential ansatz s = s0 exp(t/t0 ). The latter corresponds to the
assumption that the object moving through the nebula collects
smaller particles on its surface. A more detailed approach will
employ statistical methods (Krivov et al. 2005, 2006). Then,
we wish to explore rotation in greater detail, calculating colli-
sion timescales and rotation frequencies using models for two-
body collisions. Also, the effects of gas drag and radiation
forces on non-spherical objects need to be taken into account
(see e.g. Xu et al. 1999). Other issues include the variation of
physical parameters (density, thermal conductivity, emissivity,
J1 ) with size and distance from the star, as well as the global
evolution and clearing of dust in the system which defines the
time around which photophoresis can come into play.
In spite of these unknowns, we have demonstrated that pho-
tophoretic force in transitional circumstellar disks cannot be
neglected, and has to be included in elaborate models of such
systems.
Acknowledgments. We wish to thank Gerhard Wurm for
useful discussions, Jürgen Blum and Bo Gustafson for pointing
out several papers on the photophoretic effect and the anony-
mous referee for helpful comments. F.H. is supported by the
graduate student fellowship of the Thuringia State.
References
Alibert, Y., Mordasini, C., Benz, W., & Winisdoerffer, C. 2005,
A&A, 434, 343
Ardila, D. R., Lubow, S. H., Golimowski, D. A., et al. 2005,
ApJ, 627, 986
Arnold, S. & Lewittes, M. 1982, J. of Appl. Phys., 53, 5314
Augereau, J.-C. 2006, in: Semaine de l’Astrophysique
Francaise, ed. D. Barret, F. Casoli, G. Lagache,
A. Lecavelier, & L. Pagani, 375
Augereau, J. C., Lagrange, A. M., Mouillet, D., & Ménard, F.
1999a, A&A, 350, L51
Augereau, J.-C., Lagrange, A.-M., Mouillet, D., Papaloizou, J.
C. B., & Grorod, P. A. 1999b, A&A, 348, 557
Augereau, J. C. & Papaloizou, J. C. B. 2004, A&A, 414, 1153
Balog, Z., Muzerolle, J., Rieke, G. H., et al. 2007, ApJ, 660,
1532
Beresnev, S., Chernyak, V., & Fomyagin, G. 1993, Phys. Fluids
A, 5, 2043
Besla, G. & Wu, Y. 2007, ApJ, 665, 528
Blum, J. & Wurm, G. 2000, Icarus, 143, 138
Bouwman, J., Meyer, M. R., Kim, J. S., et al. 2006, in
Planet formation: Theory, observations and experiments, ed.
H. Klahr & W. Brandner, 14
Brauer, F., Dullemond, C. P., Johansen, A., et al. 2007, A&A,
704, 1169
Burns, J. A., Lamy, P. L., & Soter, S. 1979, Icarus, 40, 1
Calvet, N., D’Alessio, P., Watson, D. M., et al. 2005, ApJ, 630,
L185
Chen, C. H. & Kamp, I. 2004, ApJ, 602, 985
Currie, T., Balog, Z., Kenyon, S. J., et al. 2007, ApJ, 659, 599
Draine, B. T. & Weingartner, J. C. 1996, ApJ, 470, 551
Ehrenhaft, F. 1917, Phys. Z., 18, 352
Eisner, J. A., Chiang, E. I., & Hillenbrand, L. A. 2006, ApJ,
637, L133
Fisher, R. S., Telesco, C. M., Piña, R. K., Knacke, R. F., &
Wyatt, M. C. 2000, ApJ, 532, L141
Haisch, Jr., K. E., Jayawardhana, R., & Alves, J. 2005, ApJ,
627, L57
Hayashi, C., Nakazawa, K., & Nakagawa, Y. 1985, in
Protostars and Planets II, ed. D. C. Black & M. S. Matthews,
1100
Henning, T., Dullemond, C. P., Wolf, S., & Dominik, C. 2006,
in Planet formation: Theory, observations and experiments,
ed. H. Klahr & W. Brandner, 112
Hollenbach, D., Gorti, U., Meyer, M., et al. 2005, ApJ, 631,
1180
Hollenbach, D., Johnstone, D., Lizano, S., & Shu, F. 1994, ApJ,
428, 654
Hollenbach, D. J., Yorke, H. W., & Johnstone, D. 2000, in
Protostars and Planets IV, ed. V. Mannings, A. Boss, &
S. Russell (University of Arizona Press), 401
 Herrmann & Krivov: Photophoresis in transitional disks
Hueso, R. & Guillot, T. 2005, A&A, 442, 703
Jayawardhana, R., Coffey, J., Scholz, A., Brandeker, A., & van
Kerkwijk, M. H. 2006, ApJ, 648, 1206
Jayawardhana, R., Fisher, S., Hartmann, L., et al. 1998, ApJ,
503, L79
Jura, M., Malkan, M., White, R., et al. 1998, ApJ, 505, 897
Jura, M., Zuckerman, B., Becklin, E. E., & Smith, R. C. 1993,
ApJ, 418, L37
Kastner, J. H., Huenemoerder, D. P., Schulz, N. S., et al.
2001, in Young Stars Near Earth: Progress and Prospects,
ed. R. Jayawardhana & T. Greene (ASP Conf. Ser., vol. 244),
159
Klahr, H. H. & Lin, D. N. C. 2000, in Disks, Planetesimals
and Planets, ed. F. Garzón, C. Eiroa, D. de Winter, & T. J.
Mahoney (ASP Conf. Ser., vol. 219), 375
Kley, W., Papaloizou, J. C. B., & Lin, D. N. C. 1993, ApJ, 416,
679
Koerner, D. W., Ressler, M. E., Werner, M. W., & Backman,
D. E. 1998, ApJ, 503, L83
Krauss, O. & Wurm, G. 2005, ApJ, 630, 1088
Krauss, O., Wurm, G., Moussis, O., et al. 2006, A&A, 462, 997
Krivov, A. V., Löhne, T., & Sremčević, M. 2006, A&A, 455,
509
Krivov, A. V., Sremčević, M., & Spahn, F. 2005, Icarus, 174,
105
Lawson, W. A., Lyo, A.-R., & Muzerolle, J. 2004, MNRAS,
351, L39
Lissauer, J. J. 1993, ARA&A, 31, 129
Lowrance, P. J., Schneider, G., Kirkpatrick, J. D., et al. 2000,
ApJ, 541, 390
Mackowski, D. W. 1989, Int J. of Heat and Mass Transfer, 32,
843
Meyer, M. R., Backman, D. E., Weinberger, A. J., & Wyatt,
M. C. 2007, in Protostars and Planets V, ed. B. Reipurth,
D. Jewitt, & K. Keil, 573
Meyer, M. R. & Beckwith, S. V. W. 2000, in Lecture Notes in
Physics, Berlin Springer Verlag, Vol. 548, ISO Survey of a
Dusty Universe, ed. D. Lemke, M. Stickel, & K. Wilke, 341
Morfill, G. E. 1983, Icarus, 53, 41
Morfill, G. E. 1988, Icarus, 75, 371
Mouillet, D., Lagrange, A. M., Augereau, J. C., & Ménard, F.
2001, A&A, 372, L61
Paddack, S. J. & Rhee, J. W. 1975, Geophys. Res. Lett., 2, 365
Papaloizou, J. C. B. & Terquem, C. 1999, ApJ, 521, 823
Papaloizou, J. C. B., Terquem, C., & Nelson, R. P. 1999,
in Astrophysical Discs - an EC Summer School, ed. A. J.
Sellwood & J. Goodman (ASP Conf. Ser., vol. 160), 186
Rohatschek, H. 1996, J. of Aerosol Sc., 27, 467
Safronov, V. S. 1972, Evolution of the Protoplanetary Cloud
and Formation of the Earth and Planets, NASA TTF-677
Shakura, N. I. & Sunyaev, R. A. 1973, A&A, 24, 337
Shu, F. H., Adams, F. C., & Lizano, S. 1987, ARA&A, 25, 23
Simon, M. & Prato, L. 1995, ApJ, 450, 824
Stauffer, J. R., Hartmann, L. W., & Barrado y Navascues, D.
1995, ApJ, 454, 910
Steinbach, J., Blum, J., & Krause, M. 2004, Europ. Ph. J. E, 15,
287
13
Strom, S. E. & Edwards, S. 1993, in Planets Around Pulsars,
ed. J. A. Phillips, S. E. Thorsett, & S. R. Kulkarni (ASP
Conf. Ser., vol. 36), 235
Strom, S. E., Edwards, S., & Skrutskie, M. F. 1993, in
Protostars and Planets III, ed. E. H. Levy & J. I. Lunine,
837
Takeuchi, T. & Artymowicz, P. 2001, ApJ, 557, 990
Takeuchi, T., Clarke, C. J., & Lin, D. N. C. 2005, ApJ, 627, 286
Tehranian, S., Giovane, F., Blum, J., Xu, Y.-L., & Gustafson,
B. Å S. 2001, Int. J. of Heat and Mass Transfer, 44, 1649
Telesco, C. M., Fisher, R. S., Piña, R. K., et al. 2000, ApJ, 530,
329
Thébault, P. & Augereau, J.-C. 2005, A&A, 437, 141
van den Ancker, M. E., de Winter, D., & Tjin A Djie, H. R. E.
1998, A&A, 330, 145
Wahhaj, Z., Koerner, D. W., Backman, D. E., et al. 2005, ApJ,
618, 385
Weidenschilling, S. J. 1977, MNRAS, 180, 57
Weidenschilling, S. J. & Cuzzi, J. N. 1993, in Protostars and
Planets III, ed. E. H. Levy & J. I. Lunine, 1031
Weinberger, A. J., Becklin, E. E., Schneider, G., et al. 1999,
ApJ, 525, L53
Weinberger, A. J., Rich, R. M., Becklin, E. E., Zuckerman, B.,
& Matthews, K. 2000, ApJ, 544, 937
Wetherill, G. W. 1980, ARA&A, 18, 77
Wieneke, B. & Clayton, D. D. 1983, in Chondrules and their
Origins, ed. E. A. King, 284
Wurm, G. & Blum, J. 2006, in Planet formation: Theory, ob-
servations and experiments, ed. H. Klahr & W. Brandner, 90
Wurm, G. & Krauss, O. 2006, Icarus, 180, 487
Wyatt, M. C. 2005, A&A, 440, 937
Wyatt, M. C., Dermott, S. F., Telesco, C. M., et al. 1999, ApJ,
527, 918
Xu, Y.-L., Gustafson, B. Å. S., Giovane, F., Blum, J., &
Tehranian, S. 1999, Phys. Rev. E, 60, 2347
Zuckerman, B. & Becklin, E. E. 1993, ApJ, 406, L25
Zuckerman, B. & Song, I. 2004, ARA&A, 42, 685